coastal modelling with a gis … input parameters, data manipulation, and output processing...

INTRODUCTIONWater quality problems have been identified inmany coastal areas around the world. The envi-ronmental degradation of the coastal areas hasbeen attributed to municipal and industrial dis-charges. In addition to sanitary sewers andsewage treatment plant submerged discharges, anumber of storm and combined sewer outfalls(CSO) and major tributaries (creeks, rivers), havesignificant impacts on the water quality of coastalareas. During a heavy rainfall event, the runofffrom roads and buildings is collected in the com-bined storm and sanitary sewers where it is trans-ported to the sewage treatment facility. If thedemand on the plant is too high, the excess runoffand raw sewage in the sewers is redirected to out-falls located along the shoreline. In fact, it is not

uncommon to hear of beach closures in a numberof coastal areas around the world after a heavyrain event due to high pollutant levels. Significantreductions can be made to the pollutant levels ifthe redirected flow is restricted from dischargingdirectly into the coast. For example, by collectingthe overflow in large underground storage con-tainers, the level of pollutants usually contributedby the sewers is reduced to zero (Tsanis, 1996).Other potential options include separating theremaining combined systems, or decentralizingthe collection and treatment of combined sewageand storm water by adding more, smaller treat-ment facilities.Geographic Information Systems linked with dif-ferent environmental simulation models can pro-vide functions for data storage, calculation of

Global Nest: the Int. J. Vol 4, No 1, pp 51-73, 2002

Copyright 2002 GLOBAL NEST

Printed in Greece. All rights reserved

COASTAL MODELLING WITH A GIS BATHYMETRIC MODULE

Department of Civil Engineering

McMaster University

1280 Main Street West

Hamilton, Ontario, Canada L8S 4L7

*to whom all correspondence should be addressed

e-mail: [email protected]

tel: (905) 525 - 9140 ext 24415

ABSTRACTA 3D hydrodynamic/pollutant transport model was used to simulate the currents and pollutant trans-port in coastal areas. The bathymetric and shoreline data was provided to the model via a GIS modulethat operates in the ArcView GIS environment. The module is efficient and capable of generatingbathymetric rectangular grids and shorelines of different resolution for open and closed boundary sce-narios that can automatically be read by the coastal model. This ability could improve the basic patternsand relationships of the model such as grid dependency. The functionality of the 3D model with theGIS module is illustrated in a number of coastal areas in Greece.

KEY WORDS: ArcView GIS, Coastal Hydrodynamics, pollutant transport

I.K. TSANIS*

S. NAOUM

M. FULLARTON

Received: 20/11/01Accepted: 31/10/02

TSANIS-test.qxd 21/7/2003 2:05 Page 51

required input parameters, data manipulation,and output processing (Maidment, 1993; Tsanis etal., 1996; Goodchild et al., 1995; Naranjo andLarsen, 1998; Lichy, 1998). GIS technology is alsocentral to these models by providing the systemwith spatial data management, analysis, displayand interface functions (Boyle et al., 1998; Boyleand Tsanis, 1998). Hydrodynamic/pollutant trans-port modelling requires functions for data input,processing, and output of spatially distributedinformation. There are many means to providethis functionality, for example closed-coupledGIS models such as the IDOR2D (Tsanis andBoyle, 2001) and loose-coupled GIS model(Naoum et al., 2002).In this paper, a GIS interface operating in theArcView GIS environment was used to providethe bathymetic grid and shoreline needed for the3D hydrodynamic/pollutant transport modelIDOR3D, which is a DOS-based educationaland research program developed at McMasterUniversity, in order to simulate the currents andpollutant transport in a number of coastal areasin Greece. The remainder of the paper is orga-nized as follows: Section 2 provides the descrip-tion of the 3D hydrodynamic/pollutant transportmodel. In section 3 the use of the GIS module inthe hydrodynamic modelling is described.Section 4 describes the case study. In section 5the modelling results are reported and dis-cussed; and finally, section 6 concludes the pre-sentation.

METHODOLOGY

Governing EquationsThree-dimensional governing equations for wind-induced circulation in coastal areas were derivedfrom the Navier-Stokes equations by using thefollowing assumptions:

Shallow water assumption: the vertical accel-erations are small enough to be ignored, whichresults to the hydrostatic pressure distribution

Boussinesq approximation: the variation ofwater density is taken into account only in thegravity term.

(1)

(2)

(3)

(4)

(5)

(6)

where: u, v, and w are the velocity components inthe x, y, and z directions, with x positive eastward,y positive northward, and z positive upward; f isthe Coriolis parameter; p is the water pressure; gis the acceleration due to gravity; is the waterdensity; 0 is the reference water density (at4C); T is the water temperature; h and v arethe eddy viscosity coefficients in the horizontaland vertical directions, respectively; ST is thesource term for the water temperature; Kh and Kvare the eddy diffusivity coefficients for watertemperature in the horizontal and vertical direc-tions, respectively; C is the concentration; SC isthe source for concentration; and Mh and Mv arethe eddy diffusivity coefficients in the horizontaland vertical directions, respectively. Finally theequation of state = (T, S), where S is thesalinity, provides the link between density andtemperature.

2 2

2 2C h h v

C C CS M M M

x y z z

= + + +

( ) ( ) ( )uC vC wCCt x y z

+ + + =

2 2

2 2T h h v

T T TS K K K

x y z z

= + + +

( ) ( ) ( )uT vT wTTt x y z

+ + + =

z

p

1 g = 0

)z

v (

z +

y

v +

x

v +

y

p

1fu = v2

2

h2

2

h

0

= z

(wv) +

y

(vv) +

x

(uv) +

t

v

)z

u (

z +

y

u +

x

u +

x

p

1 fv = v2

2

h2

2

h

0

= z

(wu) +

y

(vu) +

x

(uu) +

t

u

0 = z

w +

y

v +

x

u

52 TSANIS et al.


Boundary ConditionsAt the water surface (z = ), the kinematic condi-tion can be derived:

(7)

where: subscript s denotes the variables at thewater surface. The wind shear stresses can be con-nected to the boundary conditions as follows:

(8)

where: W is the wind speed; a is the air densityand; s is the wind shear stress; Cs is the windshear stress coefficient. This model can incorpo-rate the heat exchange at the water surface, but itis assumed that there is no heat exchange at thewater surface in this study. No flux of concentra-tion at the water surface is considered.

At the sea bottom (z = zb ), the following kine-matic condition is obtained:

(9)

where: subscript b expresses the variables justabove the bottom boundary layer. The bottomshear stresses are given as:

(10)

where: b is the bottom shear stress and Cb is thebottom drag coefficient. No heat and concentra-tion fluxes through the sea bottom are consid-ered.

At the shore, the components of velocity perpen-dicular to the shore, and the heat and conductivi-ty fluxes through the shore are assumed to bezero. Where the shore has inflow or outflow

rivers, the velocity components normal to theshore are considered to be equal to the velocitiesof the inflow or outflow rivers, and the heat andconcentration exchanges are expressed by:

(11)

(12)

where: subscript n is the normal direction to theshore; ur, Tr, and Cr are the velocity, temperatureand concentration of the inflow or outflow rivers,respectively.At the open boundaries, the velocity and waterelevation at the open boundaries must be speci-fied. When the data of the velocity and water ele-vation at the open boundaries are not available,the following free radiation open boundary condi-tions are used.

Water elevation: (13)

Tangential velocity component:

(14)

Normal velocity component: (15)

where: subscripts n and t are the normal and tan-gential directions to the open boundary, respec-tively.

The Reynolds stress terms should be specified inorder to solve the equations of motion. Tsanis(1989) compared different turbulence models andconcluded that the mixing length model (Reid,1957), the - model (Svensson, 1978) and theeddy viscosity model (Pearce and Cooper, 1981)had very similar vertical current structures. In thispaper, because the eddy viscosity model proposedby Pearce and Cooper is much simpler than theothers, it is used for the vertical eddy viscosity. Thatis, the vertical eddy viscosity coefficient was takenas zero at the water surface, increasing linearly to 1

0 = n

un

0 = vv n

+ t

vtn

t

0 = n

Cu = n

C M Cu rrhn

Tu = n

T K Tu rrhn

v + u )v ,u( C =2b

2bbb0b

= | z

v ,

z

u = ) ,(

zz=v0bybx b

y

z v +

x

z u = w

bb

bbb

W + W )W ,W( C =2y

2xyxas

= | z

v ,

z

u = ) ,(

z=vasysx

y v +

x u +

t = w sss

53COASTAL MODELLING WITH A GIS BATHYMETRIC MODULE


metre depth from the surface, and constant fromthat depth until the bottom. This constant eddy vis-cosity is calculated by (v 0.08 u*s H), in whichu*s is the shear velocity at the water surface andH is the water depth. In stratified cases, the eddyviscosity and diffusivity are reduced by the stablestratification and often evaluated by the followingformulas (Leendertse and Liu, 1975),

(16)

where: v0, Kv0 and Mv0 are the eddy viscosity anddiffusivities in the neutral state (isothermal state)respectively, Ri is the local Richardson numberand is defined as:

(17)

in which: u is the average velocity in the horizon-tal plane. A large Richardson number impliesthat the stratification is strong and thus the verti-cal transports of momentum and pollutant aremore suppressed. If the Richardson number isnegative, turbulence is stimulated and strength-ens by the buoyancy force and the kinetic energyof turbulence will increase. As a result, the verti-cal transports of momentum and pollutant will bemore intensive than in the neutral state. In themodel, it is assumed that instantaneous mixinghappens when the Richardson number is nega-tive. This model can incorporate spatially variablehorizontal eddy viscosity h, however h was cal-culated by: h= 10

-3 h4/3 (m2 s-1) for this study, in

which h is the grid size (m) in the horizontalplane.

Numerical Analysis TechniquesMany numerical schemes have been presented fordiscretizing the governing equations. Most of thenumerical schemes are conditionally stable, thatis, the time steps are restricted by several stabilityconditions. The time step is limited by theCourant-Friedrichs-Levy (CFL) condition if thebarotropic pressure is discretized by an explicitscheme,

(18)

where: Hmax is the maximal water depth andx, y, and z are the grid lengths in the x, y,and z directions respectively. The other stabilityconditions are imposed by the diffusion

(19)

and Coriolis term: (20)

Among these stability conditions, the CFL condi-tion is the most stringent. If the CFL condition isnot imposed, a much larger time step can be cho-sen.The governing equations were approximated bythe control volume method on the Arakawa-Cstaggered grid (Arakawa, 1966). A semi-implicitdifference scheme was used for the barotropicpressure, known as the SIMPLER (Semi-ImplicitMethod for Pressure-Link Equation Revised)method (Patankar, 1980). This method results ina Poisson equation of the water elevation, whichwas solved by the SOR (Successive Over-Relaxation) method (Roache, 1972). In the SIM-PLER method, the time step is not limited by theCFL condition and thus a larger time step can bechosen to save CPU time. The Adams-Bashforthscheme was used for the temporal terms and theweight averaged Donor-cell scheme for theadvective terms. Although the Adams-Bashforthscheme has been proven by the Neumanns stabil-ity analysis to exhibit weak dispersion when onlyadvective terms exist, this scheme is stable if dif-fusive terms are involved (Roache, 1972). TheAdams-Bashforth scheme is second-order accu-rate in time, and requires values at only two timeincrements. The Donor-cell scheme is a second-order upstream scheme in space and avoidsnumerical stability problems caused by the non-linear advective terms (Roache, 1972). The cen-tral difference scheme of second order accuracyin space was used for the diffusive term. As aresult, the numerical model is second-order accu-rate in both time and space. The above-describedmodel is called IDOR3D. The procedure of cal-culation has the following steps:

The advective terms, diffusive terms, Coriolisterms and the components of pressure terms

f

1 < t

1/2 z

t +

y

1 +

x

1 t

2

v

22h

1 z

1 +

y

1 +

x

1 t gH

222

max

)z/u(

zT/g = Ri

2

0

e M = M3.0Ri

v0v

e K = K3.0Ri

v0v

e = 1.5Ri

v0v

54 TSANIS et al.


are not related to the water surface elevationn+1, are calculated from the hydraulic quanti-ties of time nt. Time is determined by thenumber of time steps nt from the referencetime, with n as an integer value.

The water surface elevation n+1 is calculatedby the Successive Over Relaxation (SOR)method from the Poisson equation, and zn+1

can be obtained from n+1.

The velocity un+1 and vn+1 are calculated fromthe momentum equations.

The vertical velocity wn+1 is calculated by sub-stituting the obtained velocities un+1 and vn+1

into the continuity equation.

The temperature T is calculated from the tem-perature equation and the density is calcu-lated from the relationship between tempera-ture and density.

The concentration C is calculated from thetransport equation.

THE GIS ROLE IN 3D HYDRODYNAMIC MODELINGA GIS module was used in this paper to prepareinput ASCII files representing the bathymetricgrid and shoreline necessary for running thehydrodynamic/pollutant transport model. Digitalhydrographic files normally represent the defini-tion of the bathymetry of the computationaldomain. Since the hydrographic files were in dec-imal degrees (geographic projection), projectingthem in the appropriate projection is the firststep in order to properly link all bays and anyadditional information. Upon completing pro-jection, the files were converted to ArcViewshapefiles; filtered to eliminate any irrelevantinformation for this specific application (i.e.rivers); edited to fix any errors and erase anyunnecessary features (i.e. moving any spot depthspoints to be inside the shoreline, trim shoreline,...etc); and manipulated to match the desired fea-tures by the module to generate grids (i.e. con-vert contours to points, convert shoreline frompolygon to polyline, merge islands and shoreline,...etc).The written scripts are then run consecutively to

perform the grid generation, which also havebeen designed to prompt the user with somedialogs and messages for guidance and interac-tion. The depth grids may be defined for either asub-area or the full extent of the available inputdata. The depth grids can also be generated foropen or closed boundary conditions as will beseen in the following sections. The automatic,quick, and accurate generation of depth grids,accompanied by the visual capability of ArcView,provides the means for studying the impact ofgrid-resolution on the 3D Hydrodyna-mic/Pollutant Transport model outputs. The usercan specify both the grid cell size and the interpo-lation technique.After collecting all necessary information fromthe user through prompts and messages, the pro-gram starts executing the task of generating thedepth grid. The execution time depends exclu-sively on the interpolation technique, the size ofthe bay, and the cell size. The user is theninformed of the completion of the operation bydisplaying a message on the screen. The ASCIIfile is then generated together with a visual pre-sentation of the grid. Finally, another process isinitiated by running a series of ArcView scriptsthat convert the shoreline (including islands ifthey exist) to points separated at user-specifiedspacing. The text file is then generated.The hydrodynamic model IDOR3D requiresadditional processing of files once the shore andbathymetric files have been generated. The fol-lowing steps are necessary for running theIDOR3D model

(a) The bottom topography of the computationaldomain must be defined. The depths of thedomain are interpolated from the topographyfile, created by ArcView GIS and an addition-al program runs to generate a depth grid.

(b)Generation of two files that contain informa-tion about each grid (column) within the larg-er grid system. The depths in each column areexpressed as cells. The number of cells in acolumn is directly related to the depth of waterat that point, and the number of layers definedin a file, which specifies the thickness of eachlayer.

(c) To define the simulation, IDOR3D reads two



parameter files. The first file providesIDOR3D with information such as simulationlength, date and time of simulation, time serieslocations, time levels when variables about thewhole water body are to be saved, as well aswind and source characteristics. The second filedefines the physical characteristics of the bodyof water provided by experts such as diffusivity,viscosity, water density, drag coefficient, etc.

CASE STUDYSelected coastal areas in Greece, as shown inFigure 1, were chosen as test cases to illustratethe functionality of coupling the 3D model withthe GIS module. In addition to the availability ofthe data, the coastal areas were selected toreflect: a) different boundary conditions, b) vari-ous sizes (areas), and c) different locationsaround the country. The data was obtained fromhydrographic files which, mostly, included infor-mation on spot depths, bathymetric contours,shorelines, islands, and rivers. Figure 2 shows anexample of a hydrographic file for the

Thessaloniki bay (map section B), which wasprocessed through the GIS module to a properformat for generating the bathymetric file.Additional bathymetric information was avail-able for the island of Crete (which is located tothe far south of Greece) and five areas wereselected that represent: (a) area E, the coast ofthe whole island; and (b) four coastal areas in thenorth shore of Crete. These cases were run forthe island employing different modeling scenar-ios that include one, two, three, and four openboundaries. Additional information from theDigital Elevation Model (DEM) of the island canprovide information, as shown in Figure 3, ofpoint sources discharging in the near shore of theisland. For complicated shorelines, small cellsizes are generally preferred in order for the gridto conform to the shore. Figure 4a shows a1000m-grid for the Amvrakikos bay (map sectionA), while Figure 4b shows a 50m-grid for thesame area. Notice the difference for the north-west part of the bay, which is enlarged on bothfigures in the insets.

56 TSANIS et al.

Figure 1. Selected bays in the country of Greece



Figure 2. A hydrographic file for the Thessaloniki bay

Figure 3. A digital bathymetry model (DBM) together with a digital elevation model (DEM) for the north

shore of the Iraklion prefecture


58 TSANIS et al.

Figure 4a. A 1000m-grid for the Amvrakikos bay

Figure 4b. A 50m-grid for the Amvrakikos bay


RESULTS AND DISCUSSIONA number of simulations were performed onvarious bays in Greece. Figure 5 shows the cur-rent distribution at the surface and at a depth of20 m based on a 5 m s-1 NW wind inThessaloniki bay (map section B) located innorth Greece. The wind-induced currents pro-

duce a consistent flow pattern at the surface andgenerate a return flow at a depth of 20 m.Figure 6 examines the surface layer current pat-terns for Kalloni bay (map section C) under aconstant 5 m s-1 wind applied from the SW andthe NW. Eddy currents form at the surfaceunder both wind conditions as a result of the


Depth (m)

Bathymetry of the Bay

34

30

26

22

18

14

10

6

2

B

20 cm/s

Current DistributionSurface Layer

Wind 5m/s

20 cm/s

Wind 5m/s

Current DistributionDepth = 20m

Figure 5. Shows the bathymetry of Thessaloniki bay (Section B) and the current distribution at the surface and

at a depth = 20 m based on a NW wind = 5 m s-1.


geometry of the bay, which is long and narrow.Figure 7 illustrates the bathymetry and surfacelayer flow distribution under a constant 5m s-1

NW wind for the Saronikos bay near Athens(map section D). The uniform depth averagedcurrent patterns are represented in Figure 7along with referenced depth averaged flow forthe NE corner of the bay. A number of eddy

currents are prevalent within the extent of thedomain due to the presence of islands, smallinlets and the bathymetry of the bay.As previously mentioned and due to the avail-ability of additional bathymetric data for theisland of Crete, five runs, employing differentboundary and pollutant source conditions, wereperformed for the island in the areas shown in

60 TSANIS et al.

Depth (m)

Bathymetry of the Bay

19

17

15

13

11

9

7

5

3

1

20 cm/s

Wind 5m/s

Current PatternsSurface LayerSW Wind = 5m/s

Wind 5m/s

20 cm/s

Current PatternsSurface LayerNWWind = 5m/s

Figure 6. Shows the bathymetry of Kalloni bay (Section C) and the current distribution based on a 5 m s-1 SW

and NW wind direction.



Depth (m)

400

375

350

325

300

275

250

225

200

175

150

125

100

75

50

25

20

15

10

5

Wind 5m/s D

30 cm/sWind 5m/sCurrent Distribution

Surface Layer

Wind 5m/s

Current DistributionDepth AveragedUniform Vectors

Inset A

10 cm/s

Inset ACurrent DistributionDepth Averaged

Wind 5m/s

Figure 7. Shows the bathymetry of Saronikos bay (Section D) the surface layer current patterns for a 5m s-1

NW wind. It also describes the depth averaged current patterns for Saronikos bay (Section D) using

uniform vectors for the entire bay and actual magnitude vectors for the NE corner of the bay.


Figure 8. The distribution of the surface layercurrents around the Island of Crete (map sectionE) under a NW wind is illustrated in Figure 9. Asection along the north shore of the island (northof the Iraklion prefecture) was also considered inorder to simulate the surface flow based on a NWand NE wind. Figure 10 describes the bathymetryand current patterns at the surface and at a depthof 200 m for Agios Nikolaos bay based on a NWwind. As expected the currents at the surface aredriven by the wind and a return flow is prevalentat a depth of 200 m following the bottom topog-raphy of the bay.The two bays located along the north shore ofChania region were used to examine the hydrody-namics and pollutant transport of a hypotheticalcontaminant discharged under various condi-tions. Kasteli bay (map section E4) was used tosimulate the fate of a pollutant discharged over afour-hour period. Chania bay (map section E3)was used to examine the transport of a pollutantdischarged continuously from a source for the

entire simulation from two different locations.The first scenario to consider involves Kasteli bay.As shown in Figure 11, a watershed is dischargingthrough two streams into the bay, which could beregarded as a source of pollution that can causeenvironmental problems in the near shore areas.Three reference points (i.e. RP1, RP2 and RP3)are positioned downstream (i.e. east) of thesource (river outlet) in order to determine theconcentration levels at these points over theentire length of the simulation. RP1 and RP2 arelocated (side-by-side) at a distance of 200 m fromthe source, which is defined as a surface dis-charge. RP3 is situated 1.5 km downstream of thesource and 200 m offshore. RP1 is positioned onthe shore and RP2 is 200 m offshore north ofRP1. The purpose of this arrangement is to com-pare the concentration levels along the shore withlevels just off shore. A NW wind is applied to thebay under isothermal conditions for a period of 72hours. A contaminant with a concentration of10000 ppm is released 24 hours after the simula-

62 TSANIS et al.

Figure 8. The DBM and DEM of the island of Crete and its selected bays



20 cm/s

Island of Crete

Current DistributionSurface Layer

E1

Wind 5m/s

Inset A

20 cm/s

Inset A (E1)Current DistributionSurface Layer

Wind 5m/s E1

Inset A (E1)Current DistributionSurface Layer

Wind 5m/s

20 cm/s

E1

Figure 9. The surface layer current patterns around the Island of Crete (Section E) and a section (E1) of the

northshore of Crete for a 5m s-1 NW wind.


64 TSANIS et al.

Depth (m)350

325

300

275

250

225

200

175

150

125

100

75

50

25

15

5Bathymetry of the Bay E2

Current DistributionSurface layer

20 cm/s

Wind 5m/s

Current DistributionDepth = 200m

5 cm/s

Wind 5m/s

Figure 10. The bathymetry of Agios Nikolaos bay (Section E2) and the current patterns at the surface layer and

at a depth of 200 m for a 5 m s-1 NW wind.


tion begins at a rate equal to 1 m3 s-1 over a four-hour period.Figure 12 shows the movement of the plume 6, 10and 20 hours after the initial release of the pollu-tant. The plume is restricted in moving along theshoreline due to the currents generated by theNW wind. The contaminant is transported down-stream as a plug flow and it takes approximatelytwenty hours (from the initial release of the pol-lutant) for the slug to pass over RP3, which is 1.5km away from the source. A comparison of thedepth-averaged time series concentration levelsat the three reference points and the source isshown in Figure 13a. Based on the results, thepollutant exhibits a higher concentration at RP1(i.e. along the shore) than at RP2, which is expect-ed due to the location of the source and thealongshore currents. Figure 13b shows the con-centration time series at the source for variouslayers. The water depth at this point is approxi-mately 5 m. Under isothermal conditions thedepth-averaged concentration is approximately400 ppm; this suggests an initial dilution of 25:1.

The second scenario involves another bay in theregion of Chania referred to as Chania bay (mapsection E3). Two cases will be used to examinethe behavior of a contaminant discharged contin-uously into the bay from a submerged source attwo different locations. Case 1: corresponds to a source positioned 2 kmoffshore and approximately 900 m NW of a smallisland. The depth of the source is 27 m (Figure14).Case 2: simulates the pollutant transport dis-charged from a source 10m deep, situated 250moffshore and 800m SW of the island (Figure 16).The purpose behind the two different orientationsof the source is to analyze the movement of thecontaminant with respect to the island. A NWwind of 5 m s-1 was applied to both cases in orderto produce similar hydrodynamic conditions in thebay. The source released a pollutant with a con-centration of 1000 ppm at a rate of 1 m3 s-1. Forboth cases three reference locations are used (i.e.RP1, RP2 and RP3). RP1 is located between themainland and the east side of the island. RP2 and


Figure 11. A watershed discharging to Kasteli bay through two streams


66 TSANIS et al.

RP1

RP2

Source

Inset A

6 hrs after the initial release

RP3

Wind 5m/sE4

Conc. ppm

120

100

80

60

40

20

18

16

14

12

10

8

6

4

2

Source

RP2

RP1

10 hrs after initial release

RP3

Wind 5m/s

E4

Source

RP2

RP120 hrs after initial release

RP3

Wind 5m/s

E4

Figure 12. Current distribution and pollutant concentration contours at the surface layer of Kasteli bay under

sea isothermal conditions and a 5 ms-1 NW wind.

Conc. ppm

120

100

80

60

40

20

18

16

14

12

10

8

6

4

2

Conc. ppm120

100

80

60

40

20

18

16

14

12

10

8

6

4

2



Figure 13. Concentration time series for a conservative pollutant released for a period of 4 hrs into Kasteli bay

under sea isothermal conditions and a 5m s-1 NW wind.

TSANIS-teliko.qxd 25/7/2003 12:46 Page 67

RP3 are defined further east, approximately 4 and7 respectively from the source and 700 m offshore.Figure 14 describes the source and referencepoint locations as well as the movement of thepollutant 12 and 48 hours after the start of thesimulation for Case 1. Due to the current pat-terns generated by the NW wind and the positionof the source, the pollutant moves eastwardaround the north side of the island and then trav-els back towards the shore. Figure 15 shows theconcentration time series at the source for Case 1at different layers under isothermal conditions.The concentration is considerably higher at thebottom layer compared to the surface layerbecause temperature is not considered underisothermal conditions therefore more mixingoccurs. A comparison of the depth-averaged con-

centration time series for the three referencepoints is shown in Figure 15b. As expected theconcentration at RP1 is lower than RP2 and RP3due to its location. Based on the position of RP1only the outer edge of the plume flows over thispoint, therefore levels will be much lower. Underisothermal conditions the concentration at RP2and RP3 reach the same constant level (equal to1.5ppm) since RP3 although is further away fromthe source as compared with RP2 but is in lowerdepth of 12 m versus 18 m. The plume move-ment for Case 2 is described in Figure 16 andsuggests that a large amount of the contaminantwill travel between the island and the mainland.In both cases the pollutant accumulates in the SEcorner of the bay based on this particular NWwind pattern.

68 TSANIS et al.


Source

RP2RP3

30 cm/s

RP1

Wind 5m/s E3

Inset A


Source

RP1

RP2

Wind 5m/s

30 cm/s

RP3 Inset A

DepthsSource = 27mRP1 = 7mRP2 = 18mRP3 = 12m

E3

Conc. ppm2

1.8

1.6

1.4

1.2

1

0.8

0.6

0.4

0.2

Figure 14. Case 1 - Current distribution and pollutant concentration contours at the surface layer for a continu-

ous conservative pollutant released with a concentration of 1000 ppm from a submerged source at a

depth of 27 m, in Chania bay under sea isothermal conditions and a 5 m s-1 NW wind.



Figure 15. Case 1 - Concentration time series for a continuous conservative pollutant released with a concentra-

tion of 1000ppm into Chania bay under sea isothermal conditions and a 5 m s-1 NW wind.


Figure 17 shows the concentration time series forthe new source location (which is located at adepth of 10 m) under both isothermal and strati-fied conditions for Case 2. The initial dilution isapproximately 22:1 and 13:1, respectively. Thedilution is less under stratified conditions due tothe buoyant forces of the pollutant generated bytemperature and density differences, whichreduces the amount of mixing due to the contam-inant moving quickly to the surface.The behavior of the contaminant downstream ofthe source is analyzed by investigating the timeseries concentrations for the three referencepoints. Figure 18 shows the concentration levelsfor RP1, RP2 and RP3 over the entire length ofthe simulation under isothermal and stratifiedconditions for Case 2. Unlike Case 1 where the

concentration at RP1 was quite low (see Figure15b), due to the new source location RP1 recordsmuch higher concentration levels than RP2 andRP3. This makes sense since most of the pollutantnow moves between the small island and theIsland of Crete and directly over RP1.

CONCLUSIONThe use of both the GIS and the 3D hydrodynam-ic/pollutant transport model, IDOR3D, proved tobe an effective and flexible tool for analysis andresearch. The GIS module, in addition to savingtime in generating input files, provides higher spa-tial accuracy since all themes are geographicallyreferenced, eliminating the possibility of overlap-ping or misplacing. It also shows great flexibility ingenerating grids of different cell sizes, interpola-

70 TSANIS et al.


Source

RP1

RP2

Wind 5m/s

30 cm/s

RP3 Inset A

DepthsSource = 10mRP1 = 7mRP2 = 18mRP3 = 12m

E3

Conc. ppm20

18

16

14

12

10

8

6

4

2

Figure 16. Case 2 - Current distribution and pollutant concentration contours at the surface layer for a continu-

ous conservative pollutant released with a concentration of 1000 ppm from a submerged source at a

depth of 10 m, in Chania bay under sea isothermal conditions and a 5m s-1 NW wind.


SourceRP2

RP3

30 cm/s

RP1

Wind 5m/s E3

Inset A



Figure 17. Case 2 - Concentration time series for a continuous conservative pollutant released with a concentra-

tion of 1000 ppm into Chania bay under a) isothermal and b) stratified conditions.

Case 2 - Source Conditions (Depth = 10 m)Isothermal Conditions

Case 2 - Source Conditions (Depth = 10 m)Stratified Conditions


72 TSANIS et al.

Figure 18. Case 2 - Depth-averaged concentration time series for a continuous conservative pollutant released

with a concentration of 1000 ppm into Chania bay under a) isothermal and b) stratified conditions.


tion techniques, and spatial extents. These ele-ments are vital for research such as trial and repe-tition in sensitivity analysis cases.The use of the module with IDOR3D was illus-trated through multi-case scenarios involving anumber of coastal areas in Greece. The differentscenarios involved different: (a) grid sizes and bay

sizes, (b) wind cases (magnitude and direction),(c) pollutant releases, and (d) open boundaryconditions. The model with the input data fromthe GIS module performed efficiently. Thesefindings are expected to contribute to improvingthe coastal engineering planning and manage-ment of the coastal areas in Greece.


REFERENCESArakawa, A. (1966), Computational design of long-term numerical integration of the equations of fluid motion.

I. Two-dimensional incompressible flow, Journal of Computational Physics, 1, 119-143.

Boyle, S.J. and Tsanis, I.K., (1998), IDOR2DGIS-A closely coupled Hydrodynamic/Pollutant Transport GIS

model, 18th Annual ESRI Int. User Conf., San Diego, California.

Boyle, S.J., Tsanis, I.K. and Kanaroglou, P.S. (1998), Developing geographic information systems for land use

impact assessment in flooding conditions, Journal of Water Resources Planning and Management,

ASCE, 124, 89-98.

Goodchild, M.F., Parks, B.O. and Steyaert, L.T. (1995), GIS and Environmental Modelling: Progress and

Research Issues, GIS World Books.

Leendertse, J.J. and Liu, S.K. (1975), A three-dimensional model for estuaries and coastal seas. Aspects of

Computation, Dept. of Interior, RAND, Santa Monica, 2, 1-29.

Lichy, C. (1998), A multidirectional and multifunctional gateway between GIS and hydrodynamic models,

Hydroinformatics 98, Copenhagen, Denmark.

Maidment, D.R. (1993), GIS and Hydrologic Modelling. Environmental Modelling with GIS. Goodchild, M.F.,

Parks, B.O., and Steyaert, L.T. (eds.), Oxford University Press, New York, N.Y., pp. 147-167.

Naoum, S., Tsanis, I.K., and Fullarton, M. (2002), A GIS Pre-Processor For Pollutant Transport Modelling,

Journal of Geographical and Environmental Modelling (to appear).

Naranjo, E., and Larsen, H., (1998), An integrated model to assess pollution loads with the use of GIS and numer-

ical models, Hydroinformatics 98, Copenhagen, Denmark.

Patankar, S.V. (1980), Numerical heat transfer and fluid flow. Hemisphere, Washington D.C.

Pearce, B.R. and Cooper, C.K. (1981), Numerical circulation model for wind-induced flow, Journal of Hydraulic

Engineering, ASCE, 107, 285-301.

Reid, R.O. (1957), Modification of the quadratic bottom-stress low of turbulent channel flow in the presence of

surface wind-stress. Tech. Memor. No. 93, Beach Erosion Board, U.S. Dept. Army, Washington DC.

Roache, P.J. (1972), Computational fluid dynamics, Hermosa Publishers, Albuquerque, New Mexico.

Svensson, U. (1978), Mathematical model of the seasonal thermocline, Report No. 1002, Department of Water

Resources Engineering, Univ. of Lund, Sweden

Tsanis, I.K. (1989), Simulation of wind-induced water currents, Journal of Hydraulic Engineering, ASCE, 115,

1113-1134.

Tsanis, I.K. (1996), Metropolitan Toronto Receiving Water Simulation Model, Protection and Restoration of the

Environment III, Proceedings of an International Conference, Chania, Greece,0 48-56.

Tsanis, I.K. and Boyle, S. (2001), A 2D Hydrodynamic/Pollutant Transport GIS Model, Journal of Advances in

Engineering Software, 32, 353-361.

Tsanis, I.K., Valeo, C., Wu, J., and Boyle, S. (1996), Managing contaminated sediments using a hydrodynamic

model and a GIS, Environmental Technology, 17, 877-883.


coastal modelling with a gis … input parameters, data manipulation, and output processing...

Documents